grumbel

24 ms·

271.
by grumbel 3y ago
I can see the point in Lanier argument, however there is a much bigger problem for VR: There is no VR analog to Reddit, Instagram, TikTok, Youtube and Co. VR is still very much an activity in which you have to actively engage, you have to f
272.
by grumbel 3y ago
> there are 2.5 billion adult gamers in the world. I don't think casually playing a mobile game every once in a while translates to going full VR. The amount of gamers that care enough about the medium to buy a console or gaming PC
273.
by grumbel 3y ago
The issue with that is that Meta doesn't really want a younger audience. Quest2 is already extremely popular with kids, but Meta's "age 13" requirement for accounts doesn't allow many of them on their platform. The
274.
by grumbel 3y ago
That's solvable, just add proper stereo pass-through cameras to turn the VR headset into an AR headset. The Lenovo Mirage Solo[1] had that four years ago, due to the death of the whole Daydream platform it didn't see much use outs
275.
by grumbel 3y ago
It doesn't need to be lossless to be useful. A lossy LLM can just throw all the redundant stuff away and focus on the actual information in the text, instead of the text word by word. It wouldn't need to keep dozens of languages a
276.
by grumbel 3y ago
There is an old EEVblog[1] video on the topic. Mine doesn't do it from chairs, but it regularly blacks out if I turn on the nearby LED strip. [1] https://www.youtube.com/watch?v=r-V_Z3bD_PA
277.
by grumbel 3y ago
CLIP Interrogator[1], which is also build into AUTOMATIC1111, gives quite reasonable results, at least if all you need is a prompt, it can't handle complex interactions: Image: https://i.imgur.com/husplYZ.png Output: &
278.
by grumbel 3y ago
They don't have to store it, they can delete it after account creation or just keep a hash in case they want to avoid account farming. Youtube is already collecting IDs in the EU if you want to view age-restricted content. Some countri
279.
by grumbel 3y ago
While it's true that you need to feed some source images into the AI to get good results, what you feed into them can be pretty arbitrary and low quality, as all you are really doing is providing the AI with information about positioni
280.
by grumbel 3y ago
You don't need to predict if it what written by LLM, if it's a human or machine makes no difference to the validity of a text. You just need to be able to extract the actual information out of it and cross check it against other s
281.
by grumbel 3y ago
> You click the channel's name, you click "videos", and you scroll down, that's it. It's an infinite-scroll list, it loads in 9 or so videos at a time, than it takes a second to update your scrollbar before you c
282.
by grumbel 3y ago
I am more optimistic here. While LLMs allow you to produce tons of garbage, they also provide the tools to filter through that garbage, something we didn't have before. LLMs allow us to view content in a way that we decided, not the co
283.
by grumbel 3y ago
The main problem with Youtube, and almost every other social site, is not the algorithm itself, but that the algorithm is the only way you can interact with that site. The moment you want to do something different, it just doesn't work
284.
by grumbel 3y ago
I'd much rather have one that tried to be objective, than just one that embraces their bias. The later one will always be terrible at reporting. The former at least has a chance of being good. Seriously, the bar is not exactly high. St
285.
by grumbel 3y ago
"Unique content" is half an hour away of getting absorbed into whatever AI model you have, which can than crunch out an endless amount of variations on it. The issue with AI art isn't just what it can do, but how insane
286.
by grumbel 3y ago
I think that depends on intend. If you train the model specifically to clone an artists artstyle or characters, maybe even use their name for marketing (in the style of <famous artist>), than that might be a copyright violation and&#x
287.
by grumbel 4y ago
> and think you need to put in an "art" image to get an "art" image out, but of course that's not While you don't need that to generate an image, it's something SD can actual do extremely well with Cont
288.
by grumbel 4y ago
> That to me seems to be quite a badly written law. It's pretty normal for a law to not be specific on the technicalities so they don't have to update the law whenever the software changes. The de facto standard to prevent bots
289.
by grumbel 4y ago
It's right there in the law: "A reservation of use in the case of works which are available online is effective only if it is made in a machine-readable format."
290.
by grumbel 4y ago
LAION is a German company and what StableDiffusion is doing seems to be covered under UrhG § 44b. If artists don't want their work inspected by bots they have the option to put a robots.txt on their site. https://www.gesetze
291.
by grumbel 4y ago
> copyright and license laundering The nice thing is that this works in both directions. Just as Microsoft and Co. can train on people's data, nothing stops the people from training on Microsoft's chatbot output. As long as we
292.
by grumbel 4y ago
None of that is accessible in the normal user interface, nor is there an easy way to disable any of that stuff from happening in the first place. As said, we used to have buttons to disable Javascript, CSS, image loading and cookies straigh
293.
by grumbel 4y ago
The Javascript issue never made sense to me. You are reading a non-free document to begin with, what difference does it make if it is running Javascript or not? If they wanna spy on you they can do that with HTML and CSS just as well. Even
294.
by grumbel 4y ago
IPFS is a HTTP replacement much more than it is a BT replacement. Meaning it provides namespace to access and link content, but unlike HTTP that is build around server addresses, IPFS build around content addressing. Another big advantage I
295.
by grumbel 4y ago
The thing that would make a Metaverse interesting is if it acted as glue between all those very similar services, i.e. being able to hop from one MMO into another one with your complete party, without having to switch accounts, avatars and
296.
by grumbel 4y ago
All the Oculus hype was based on the promise of a $300 headset. The amount of people that are willing to pay $500-$2k is rather tiny (<< 1 million), especially with a near complete lack of AAA game content. Meanwhile the VR-as-monitor
297.
by grumbel 4y ago
It's easy to see why the motion controllers we had failed, but it's hard to tell if a better ones would have been successful. The original Wiimote only had an accelerometer, that worked well enough in a game like WiiSports where a
298.
by grumbel 4y ago
Given how things have been going, I don't think VR will hit mainstream until the resolution (PPD) of the headsets can compete with a TV/monitor. One of the main problem with VR right now is that it cuts you off from all the 2D con
299.
by grumbel 4y ago
Predicting the next token perfectly would make a chatbot an omniscient god, as it could tell you things like the lottery numbers from next week. Chatbots "just" being a token predictors does not limit their power or intelligence i
300.
by grumbel 4y ago
Simple models will very quickly produce content that doesn't fit. Just stringing a bit of grammar and a dictionary together will produce locally correct sentences, but it will completely fail at generate larger texts with context. GPT

More ›